How can I check a GEO vendor's published measurement method?
A published method is useful when you can reconstruct a reported figure and identify its limits. Dentrix's 22 September 2026 record retained 149 valid answers after one exclusion from 150 planned observations, with 120 naming the vendor. Its 80.5 percent naming rate needs those definitions and counts, not just the displayed percentage.
What we measured
Measured. The September Index method asked ten fixed buyer questions per category, three times on ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode. Dentrix's naming count is 120 of 149, with a Wilson 95 percent interval of 73.4 to 86.1 percent. In a separate construction question captured on the same date, Procore was named in zero of three AI Overviews answers and three of three AI Mode answers. The examples show event, exclusion and engine granularity; they do not test another supplier's implementation or establish a stable difference from three observations.
What this means for you
Reasoned. Pick one claim in the sample report and follow its documentation. The method should tell you which records enter the count and how another reader could reach the same number.
| Method field | What it must resolve |
|---|---|
| Event | What counts as naming, citation or another reported outcome. |
| Population | Which questions, category and dated collection are included. |
| Engines | Which separately identified surfaces were observed. |
| Repeats | How repeated observations enter the result. |
| Exclusions | Why a planned observation did not enter the valid base. |
| Evidence | Where the retained answer and calculation can be inspected. |
Use the Dentrix example to test denominator wording: 149 valid is different from 150 planned. Use the Procore pair to test engine wording: a combined Google label cannot preserve a difference between AI Overviews and AI Mode.
Then compare the documented method with the sample output. A report may omit a field that the method promises to retain. Ask for the missing record before interpreting a change or buying a longer series. If the evidence cannot resolve the claim, record that limitation instead of guessing how the supplier collected it.
This checks the transparency of the proposed evidence. It does not make the supplier's score comparable to another score until their events, bases and conditions match.
- For a buyer: recompute one result before accepting the method as sufficient.
- For an agency: preserve the version of the method used for the client report.
- For a supplier: show exclusions and source records with the calculated figure.
Also asked as
How can I check a GEO vendor's published measurement method?
Related
- Dentrix's denominator example
- The Procore engine comparison
- What a valid measurement needs
- Check required engine coverage
- Compare agency evidence
Next step
Use the Free 10-question check (one engine). No email is needed; runs are limited per network and per day, and by the site's daily allowance. Keep the check's event and single-engine scope beside any comparison.
When that review identifies one supported source-page change, Fix one answer, $490.
Every figure comes from the published method, v1.1.
Claiming or correcting a vendor record is free and never depends on a purchase.
Prepared by the Broadcastwell team.
Page owner: Priya Nair, Management Consultant. Corrections: hello@broadcastwell.com